September 8, 2008

TECHNICAL SUPPORT DOCUMENT FOR THE
PETROLEUM REFINING SECTOR: PROPOSED
RULE FOR MANDATORY REPORTING OF
GREENHOUSE GASES

Office of Air and Radiation
U.S. Environmental Protection Agency

September 8, 2008


-------
CONTENTS

1.	Industry Description	3

2.	Total Emissions	11

3.	Emission Sources	12

3.1	Stationary Combustion Sources	12

3.2	Flares	13

3.3	Catalytic Cracking Units	14

3.4	Catalytic Reforming Units	14

3.5	Fluid Coking Units	15

3.6	Sulfur Recovery Vents	15

3.7	Hydrogen Plants	15

3.8	Fugitive Emission Sources	15

4.	Review of Existing Programs and Methodologies	16

5.	Types of Information to be Reported	16

5.1	Types of Emissions to be Reported	16

5.2	Other Information to be Reported	18

6.	Options for Reporting Threshold	19

7.	Options for Monitoring Methods	19

7.1	Stationary Combustion Sources	19

7.2	Flares	20

7.3	Process Emissions	20

7.4	Fugitive Emission Sources	21

8.	Options for Estimating Missing Data	22

9.	QA/QC Requirements	22

10.	References	23

Appendix A. Derivation of Calculation Methods	25

A. 1 Equipment Leaks	25

A. 2 Storage Tanks	29

A. 3 Delayed Coking Unit Coke Cutting	31

2


-------
1. Industry Description

Petroleum refineries are facilities that are engaged in producing liquefied petroleum gases
(LPG), motor gasoline, jet fuels, kerosene, distillate fuel oils, residual fuel oils, lubricants,
asphalt (bitumen), and other products through distillation of crude oil or through redistillation,
cracking, or reforming of unfinished petroleum derivatives. There are three basic types of
refineries: topping, hydro skimming, and upgrading (also referred to as "conversion" or
"complex"). Topping refineries have a crude distillation column and produce naphtha and other
intermediate products, but not gasoline. There are only a few topping refineries in the U.S.,
predominately in Alaska. Hydroskimming refineries have mild conversion units such as
hydrotreating units and/or reforming units to produce finished gasoline products, but they do not
upgrade heavier components of the crude oil that exit near the bottom of the crude distillation
column. Some topping/hydro skimming refineries specialize in processing heavy crude oils to
produce asphalt. There are 8 operating asphalt plants and approximately 20 other
hydroskimming refineries operating in the U.S. as of January 2006 (EIA, 2006a). The vast
majority (approximately 75 to 80 percent) of U.S. refineries are upgrading/conversion refineries.
Upgrading/conversion refineries have cracking or coking operations to convert long-chain, high
molecular weight hydrocarbons ("heavy distillates") into smaller hydrocarbons that can be used
to produce gasoline product ("light distillates") and other higher value products and
petrochemical feedstocks. The U.S., Western Europe, and Asia are the largest and most
sophisticated producers of refined petroleum products. The U.S. produced 23 percent of the
world's refinery products in 2003 (U.S. DOE, 2007).

There are 150 petroleum refineries in 35 States and 2 U.S. territories, eight of which are
considered asphalt refineries. The majority of oil distillation capacity in the U.S. is at large,
integrated companies with multiple refining facilities. About 30 percent of all facilities are small
operations producing fewer than 50,000 barrels per day, representing about 5 percent of the total
output of petroleum products annually. As of January 2006, the combined operating crude
capacity for the 150 U.S. refineries (including those in U.S. territories) was 17.9-million barrels
per calendar day (bbls/cd). The combined operating crude capacity for the 148 refineries within
the U.S. States is just over 17.3-million bbls/cd; the 2 refineries located in U.S. territories had a
combined capacity of 572,900 bbls/cd (EIA, 2006b). The average crude capacity utilization rate
for U.S refineries was 90.6 percent in 2005. In 2005, the U.S. refineries (not including U.S.
territories) produced 2,225 million barrels (MMbbl) of motor gasoline, 1,443 MMbbl of diesel
fuel oil, 564 MMbbl of jet fuel, 305 MMbbl of petroleum coke, 229 MMbl of residual fuel oil,
and 209 MMbbl of LPG, as well as other products (EIA, 2006a). Table 1 includes a list of the
150 refineries within the U.S. and its territories along with crude oil distillation capacities and
process charge or production capacities for various process units as of January 2006.

Petroleum refining is a very energy-intensive industry. In 2002, it accounted for about 7 percent
of the total U.S. energy consumption, making it the nation's second-highest industrial consumer
(U.S. DOE, 2007).

3


-------
Table 1. Petroleum Refineries in the United States.3

Facility Name

City

State

Crude
Capacity
(bbl/cd)

Charge capacity (bbl/sd)

Production capacity

FCCU

CRU

Fluid
Coking

Delayed
Coking

Hydrogen
(MMcfd)

Sulfur
(short
tons/d)

BP

Prudhoe Bay

AK

12,500

0

0

0

0

0

0

ConocoPhillips Alaska

Kuparuk

AK

14,000

0

0

0

0

0

0

Flint Hills Resources

North Pole

AK

210,000

0

0

0

0

0

0

Petro Star Inc.

North Pole

AK

17,000

0

0

0

0

0

0

Petro Star Inc.

Valdez

AK

48,000

0

0

0

0

0

0

Tesoro Petroleum Corp.

Kenai

AK

72,000

0

13,000

0

0

13

20

Goodway Refining LLC

Atmore

AL

4,100

0

0

0

0

0

0

Gulf Atlantic Operations LLC

Mobile Bay

AL

16,700

0

0

0

0

0

0

Hunt Refining Co.

Tuscaloosa

AL

34,500

0

7,200

0

14,000

6

80

Shell Chemical LP

Saraland

AL

80,000

0

20,000

0

0

0

35

Cross Oil & Refining Co. Inc.

Smackover

AR

7,200

0

0

0

0

3

0

Lion Oil Co.

El Dorado

AR

70,000

19,900

14,800

0

0

0

157

Big West of CA

Bakersfield

CA

66,000

0

16,300

0

22,000

25

105

BP

Carson

CA

260,000

102,500

52,000

0

65,000

105

350

Chevron USA Inc.

El Segundo

CA

260,000

74,000

49,000

0

66,000

77

600

Chevron USA Inc.

Richmond

CA

242,901

90,000

71,300

0

0

181

789

ConocoPhillips

LA - Carson/
Wilmington

CA

139,000

50,280

36,750

0

52,200

105

370

ConocoPhillips

SF - Rodeo

CA

76,000

0

32,000

0

27,000

84

310

ConocoPhillips

Arroyo Grande
(Santa Maria)

CA

44,200

0

0

0

23,400

0

120

Edgington Oil Co.

Long Beach

CA

26,000

0

0

0

0

0

0

ExxonMobil Corp.

Torrance

CA

149,500

100,000

20,000

0

54,600

138

400

Greka Energy

Santa Maria

CA

9,500

0

0

0

0

0

0

4


-------
Table 1. Petroleum Refineries in the United States.3

Facility Name

City

State

Crude
Capacity
(bbl/cd)

Charge capacity (bbl/sd)

Production capacity

FCCU

CRU

Fluid
Coking

Delayed
Coking

Hydrogen
(MMcfd)

Sulfur
(short
tons/d)

Kern Oil & Refining Co.

Bakersfield

CA

26,000

0

3,300

0

0

0

5

Lunday-Thagard Co. (aka World
Oil Co.)

South Gate

CA

8,500

0

0

0

0

0

0

Paramount Petroleum Corp.

Paramount

CA

50,000

0

8,500

0

0

0

40

San Joaquin Refining Co., Inc.

Bakersfield

CA

15,000

0

0

0

0

4

3

Shell Oil Products US

Martinez

CA

155,600

73,000

31,000

22,500

27,500

107

364

Shell Oil Products US

Wilmington

CA

98,500

36,000

34,000

0

40,000

15

280

Ten By Inc.

Oxnard

CA

2,800

0

0

0

0

0

0

Tesoro

Golden Eagle

CA

166,000

71,000

42,000

48,000

0

82

200

Valero Energy (Ultramar, Inc.)

Wilmington

CA

80,887

52,000

17,000

0

29,000

0

230

Valero Energy Corp.

Benicia

CA

144,000

75,300

37,200

29,500

0

141

303

Valero Energy Corp.

Wilmington

CA

6,200

0

0

0

0

0

0

Suncor Energy

Commerce City

CO

62,000

20,000

10,500

0

0

0

106

Suncor Energy

Denver

CO

32,000

10,000

10,000

0

0

0

2

Valero Energy Corp.

Delaware City

DE

181,500

87,000

43,800

53,000

0

40

596

Citgo Petroleum

Savannah

GA

28,000

0

0

0

0

0

0

Chevron USA Inc.

Honolulu
(Barber's Point)

HI

54,000

22,000

0

0

0

3

0

Tesoro Hawaii Petrol.

Kapolei

HI

93,500

0

13,000

0

0

18

34

ConocoPhillips

Wood River

IL

306,000

101,000

96,500

0

18,000

57

504

ExxonMobil Corp.

Joliet

IL

238,500

98,000

52,200

0

59,100

0

660

Marathon Petroleum Co. LLC

Robinson

IL

192,000

49,000

74,500

0

29,400

0

202

PDV Midwest Refining

Lemont

IL

167,000

70,000

31,200

0

44,000

12

400

BP

Whiting

IN

410,000

169,000

90,000

0

36,000

31

550

5


-------
Table 1. Petroleum Refineries in the United States.3

Facility Name

City

State

Crude
Capacity
(bbl/cd)

Charge capacity (bbl/sd)

Production capacity

FCCU

CRU

Fluid
Coking

Delayed
Coking

Hydrogen
(MMcfd)

Sulfur
(short
tons/d)

Countrymark Cooperative, Inc.

Mt. Vernon

IN

23,000

8,400

6,500

0

0

0

9

Coffeyville Refining

Coffeyville

KS

112,000

32,500

17,000

0

19,500

0

146

Frontier Oil Corp.

El Dorado

KS

106,000

39,000

30,000

0

19,000

6

230

National Cooperative Refinery
Association

McPherson

KS

81,200

24,500

23,500

0

22,000

0

81

Marathon Petroleum Co. LLC

Catlettsburg

KY

222,000

99,000

50,000

0

0

0

448

Somerset Refinery Inc.

Somerset

KY

5,500

0

1,000

0

0

0

0

Calcasieu Refining Co.

Lake Charles

LA

30,000

0

0

0

0

0

0

Calumet Lubricants Co.

Shreveport

LA

42,000

10,500

8,000

0

0

6

10

Calumet Lubricants Co.

Cotton Valley

LA

13,020

0

0

0

0

2

0

Calumet Lubricants Co.

Princeton

LA

8,300

0

0

0

0

5

3

Citgo Petroleum Corp.

Lake Charles

LA

429,500

150,000

110,800

0

104,000

0

640

ConocoPhillips

Westlake

LA

239,400

50,000

44,000

0

64,000

0

860

ConocoPhillips

Belle Chasse

LA

247,000

104,000

44,600

0

27,000

0

125

ExxonMobil Corp.

Baton Rouge

LA

501,000

241,000

78,000

0

118,500

0

800

ExxonMobil Corp.- Chalmette

Chalmette

LA

188,160

71,600

49,400

0

35,000

0

935

Marathon Petroleum Co. LLC

Garyville

LA

245,000

131,000

48,500

0

37,400

0

790

Motiva Enterprises

Norco

LA

226,500

114,000

62,000

0

23,600

60

169

Motiva Enterprises

Convent

LA

235,000

92,000

40,000

0

0

63

728

Murphy Oil USA Inc.

Meraux

LA

120,000

37,000

32,000

0

0

0

31

Pelican Refining Co. LLC

Lake Charles

LA

0

0

0

0

0

0

0

Placid Refining Inc.

Port Allen

LA

56,000

20,500

11,000

0

0

0

28

Shell Chemical Co.

St. Rose

LA

55,000

0

0

0

0

0

0

Valero Energy Corp.

Krotz Springs

LA

80,000

34,000

13,000

0

0

0

0

6


-------
Table 1. Petroleum Refineries in the United States.3

Facility Name

City

State

Crude
Capacity
(bbl/cd)

Charge capacity (bbl/sd)

Production capacity

FCCU

CRU

Fluid
Coking

Delayed
Coking

Hydrogen
(MMcfd)

Sulfur
(short
tons/d)

Valero Energy Corp.

Norco

LA

185,003

105,380

25,000

0

70,400

0

393

Marathon Petroleum Co. LLC

Detroit

Ml

100,000

30,000

20,000

0

0

0

147

Flint Hills Resources

Rosemount

MN

279,300

86,500

51,300

0

70,800

115

999

Marathon Petroleum Co. LLC

St. Paul Park

MN

70,000

26,000

20,500

0

0

9

112

Chevron USA Inc.

Pascagoula

MS

330,000

67,000

96,000

0

105,000

230

1,300

Ergon Refining Inc.

Vicksburg

MS

23,000

0

0

0

0

8

0

Hunt Southland Refining

Sandersville

MS

11,000

0

0

0

0

0

0

Cenex Harvest States

Laurel

MT

55,000

13,500

12,000

0

0

30

130

ConocoPhillips

Billings

MT

58,000

21,490

13,550

0

19,950

20

246

ExxonMobil Corp.

Billings

MT

60,000

23,500

12,500

10,400



24

0

Montana Refining Co.

Great Falls

MT

8,200

2,500

1,030

0

0

2

0

Tesoro

Mandan

ND

58,000

30,600

12,100

0

0

0

17

Chevron USA

Perth Amboy

NJ

80,000

0

0

0

0

0

0

Citgo Asphalt Refining Co.

Paulsboro

NJ

32,000

0

0

0

0

0

0

ConocoPhillips

Linden

NJ

238,000

145,000

32,000

0

0

10

180

Hess Corporation

Port Reading

NJ

0

65,000

0

0

0

0

10

Sunoco, Inc.

Westville

NJ

145,000

57,000

30,000

0

0

0

80

Valero Energy Corp.

Paulsboro

NJ

160,000

55,000

30,000

0

27,000

9

206

Giant Refining Co.

Bloomfield

NM

16,800

6,500

4,000

0

0

0

3

Giant Refining Co.

Gallup

NM

20,800

11,500

6,800

0

0

0

2

Navajo Refining Co.

Artesia +
Lovington

NM

75,000

27,000

18,000

0

0

0

130

Foreland Refining Co.

Tonopah/ Eagle
Springs

NV

2,000

0

0

0

0

0

0

7


-------
Table 1. Petroleum Refineries in the United States.3

Facility Name

City

State

Crude
Capacity
(bbl/cd)

Charge capacity (bbl/sd)

Production capacity

FCCU

CRU

Fluid
Coking

Delayed
Coking

Hydrogen
(MMcfd)

Sulfur
(short
tons/d)

BP PLC

Toledo

OH

131,000

55,000

43,000

0

34,000

33

340

Marathon Petroleum Co. LLC

Canton

OH

73,000

25,000

19,000

0

0

0

110

Sunoco, Inc.

Toledo

OH

160,000

79,000

48,000

0

0

45

62

Valero Energy Corp.

Lima

OH

146,900

40,000

55,000

0

23,000

58

110

ConocoPhillips

Ponca City

OK

194,000

69,802

54,190

0

27,660

35

34

Sinclair Oil Corp.

Tulsa

OK

70,300

27,750

16,800

0

0

0

28

Sunoco, Inc.

Tulsa

OK

85,000

0

24,000

0

11,000

0

0

Valero Energy Corp.

Ardmore

OK

83,640

30,000

21,350

0

0

26

243

Wynnewood Refining Co.

Wynnewood

OK

54,000

21,000

15,000

0

0

0

36

Chevron USA

Portland

OR

0

0

0

0

0

0

0

American Refining Group

Bradford

PA

10,000

0

1800

0

0

0

0

ConocoPhillips

Trainer
(Marcus Hook)

PA

185,000

53,000

50,000

0

0

0

41

Sunoco, Inc.

Marcus Hook

PA

175,000

105,000

20,000

0

0

0

33

Sunoco, Inc. (combined Sun &
Chevron)

Phil. (Girard Pt
& Pt Breeze)

PA

335,000

123,500

86,000

0

0

0

260

United Refining Co.

Warren

PA

65,000

26,000

14,000

0

0

0

70

Valero Energy Corp.

Memphis

TN

180,000

70,000

36,000

0

0

0

116

AGE Refining & Manufacturing

San Antonio

TX

12,200

0

0

0

0

0

0

Alon USA Energy Inc.

Big Spring

TX

67,000

25,000

21,000

0

0

0

150

BP

Texas City

TX

437,000

189,300

138,000

0

43,000

0

1,400

Citgo

Corpus Christi

TX

156,000

81,800

52,500

0

43,500

0

357

ConocoPhillips

Borger

TX

146,000

72,300

26,900

0

0

91

340

ConocoPhillips

Sweeny

TX

247,000

119,000

37,500

0

74,100

155

595

8


-------
Table 1. Petroleum Refineries in the United States.3

Facility Name

City

State

Crude
Capacity
(bbl/cd)

Charge capacity (bbl/sd)

Production capacity

FCCU

CRU

Fluid
Coking

Delayed
Coking

Hydrogen
(MMcfd)

Sulfur
(short
tons/d)

Delek Refining Ltd

Tyler

TX

58,000

20,250

17,500

0

6,000

0

15

ExxonMobil Corp.

Beaumont

TX

348,500

117,700

155,000

0

50,700

55

636

ExxonMobil Corp.

Baytown

TX

562,500

221,000

126,000

42,000

46,500

0

1,796

Flint Hills Resources

Corpus Christi

TX

288,126

106,700

71,600

0

14,000

0

237

Lyondell-Citgo Refining Co.

Houston

TX

270,200

100,000

37,000

0

105,000

0

803

Marathon Petroleum Co. LLC

Texas City

TX

72,000

55,000

11,000

0

0

0

0

Motiva Enterprises

Port Arthur

TX

285,000

90,000

48,000

0

57,500

0

711

Pasadena Refining Systems Inc.

Pasadena

TX

100,000

56,000

23,000

0

12,500

0

28

Shell Oil Products US - Deer Park
Refining Limited Partnership

Deer Park

TX

333,700

75,000

72,000

0

88,000

108

1,150

South Hampton Resources Inc.

Silsbee

TX

0

0

1,500

0

0

2

0

Total SA

Port Arthur

TX

232,000

75,000

39,600

0

0

0

300

Trigeant Ltd.

Corpus Christi

TX

0

0

0

0

0

0

0

Valero Energy Corp.

Corpus Christi

TX

142,000

114,500

69,000

0

18,500

195

1,288

Valero Energy Corp.

Houston

TX

83,000

65,000

11,500

0

0

0

110

Valero Energy Corp.

Texas City

TX

213,750

83,000

16,500

0

50,000

0

924

Valero Energy Corp.

Three Rivers

TX

90,000

24,000

33,000

0

0

12

62

Valero Energy Corp.

Sunray

TX

158,327

54,465

47,400

0

0

0

60

Valero Energy Corp.

Port Arthur

TX

260,000

78,600

53,800

0

103,000

0

1,197

Western Refining

El Paso

TX

116,000

34,000

25,000

0

0

0

40

Big West Oil Co.

Salt Lake City

UT

29,400

11,000

7,300

0

0

0

4

Chevron USA

Salt Lake City

UT

45,000

14,000

8,000

0

8,500

0

21

Holly Corp.

Woods Cross

UT

24,700

8,900

7,700

0

0

0

10

9


-------
Table 1. Petroleum Refineries in the United States.3

Facility Name

City

State

Crude
Capacity
(bbl/cd)

Charge capacity (bbl/sd)

Production capacity

FCCU

CRU

Fluid
Coking

Delayed
Coking

Hydrogen
(MMcfd)

Sulfur
(short
tons/d)

Silver Eagle Refining Inc.

Woods Cross

UT

10,250

0

2,200

0

0

1

0

Tesoro

Salt Lake City

UT

58,000

25,200

12,600

0

0

0

18

Giant Refining

Yorktown

VA

58,600

30,200

12,100

0

19,000

0

39

BP

Ferndale

WA

225,000

0

63,000

0

64,000

128

242

ConocoPhillips

Ferndale

WA

96,000

33,500

17,400

0

0

0

55

Shell Oil Products US

Anacortes

WA

145,000

57,900

32,700

0

25,700

0

350

Tesoro

Anacortes

WA

120,000

48,000

26,000

0

0

0

0

US Oil & Refining Co.

Tacoma

WA

37,850

0

6,500

0

0

0

10

Murphy Oil USA Inc.

Superior

Wl

34,300

11,000

8,000

0

0

0

34

Ergon-West Virginia Inc.

Newell (Congo)

WV

20,000

0

3,400

0

0

1

1

Frontier Oil & Refining Co.

Cheyenne

WY

47,000

12,000

9,200

0

10,000

6

101

Little America Refining Co.

Evansville
(Casper)

WY

24,500

11,000

6,000

0

0

0

0

Silver Eagle Refining Inc.

Evanston

WY

3,000

0

2,150

0

0

0

0

Sinclair Oil Corp.

Sinclair

WY

66,000

21,806

12,500

0

0

26

47

Wyoming Refining Co.

Newcastle

WY

12,500

5,500

2,750

0

0

0

4

Total in US States





17,333,014

6,275,123

3,859,070

205,400

2,305,510

2,823

32,421

Shell Chemical Yabucoa Inc.

Yabucoa

P.Rico

77,900

0

20,000

0

0

0

22

Hovensa LLC

Kingshill
(St. Croix)

V.lsl

495,000

150,000

115,000

0

61,000

0

550

Grand Total





17,905,914

6,425,123

3,994,070

205,400

2,366,510

2,823

32,993

a EIA, 2006. Tables 3 and 4. Abbreviation for capacity units are: bbl/cd = barrel per calendar day = maximum capacity considering scheduled maintenance over
365 calendar days; bbl/cd = barrel per stream day = maximum capacity for a single, operating day; MMcfd = million cubic feet per day.

10


-------
2. Total Emissions

The petroleum refining industry is the nation's second-highest industrial consumer of energy.
Nearly all of the energy consumed is fossil fuel for combustion; therefore, the petroleum refining
industry is a significant source of greenhouse gas (GHG) emissions. In addition to the
combustion-related sources (e.g., process heaters and boilers), there are certain processes, such
as fluid catalytic cracking units (FCCU), hydrogen production units, and sulfur recovery plants,
that have significant process emissions of carbon dioxide (CO2). Methane (CH4) emissions from
a typical petroleum refinery arise from process equipment leaks, crude oil storage tanks, asphalt
blowing, and delayed coking units. System blow down and flaring of waste gas also contributes
to the overall CO2 and CH4 emissions at the refinery. Additional detail on process-specific
sources of GHG emissions is provided in Section 3.

Because much of the GHG emissions from petroleum refineries are characterized as fossil fuel
combustion in the overall energy sector of the Inventory of U.S. Greenhouse Gas Emissions and
Sinks, it is difficult to obtain a clear GHG emission estimate for petroleum refineries from the
U.S. Inventory (EPA, 2008). Therefore, a separate industry profile was used to estimate the total
GHG emissions from petroleum refineries (Coburn, 2007). This profile included estimates of
CO2, CH4, and nitrous oxide (N2O) from combustion sources (including coke combustion in
FCCU), flares, and fugitive emission sources (equipment leaks, storage tanks, and wastewater
treatment). The profile was updated to include GHG emissions from hydrogen production units,
sulfur recovery units, asphalt blowing operations, and blowdown systems using the following
emission factors:

¦	Hydrogen production: 6.05 metric tons CCVmillion cubic feet hydrogen produced

¦	Sulfur production: 0.366 metric tons CCVshort ton sulfur produced

¦	Asphalt blowing: 2,555 standard cubic feet CH4/thousand barrels asphalt blown

¦	Blow down systems: 137 standard cubic feet CH4/thousand barrels crude throughput

The hydrogen production emission factor was based on the mass of CO2 produced being 2.5
times the mass on hydrogen produced as reported in the "Hydrogen Fact Sheet" (see:
http://www.getenergvsmart.org/Files/HvdrogenEducation/6HvdrogenProductionSteamMethaneR
eforming.pdf). The sulfur production emission factor was based on engineering estimate after
evaluating tail gas flow rates for sulfur recovery plants reported in EPA, 1998. The asphalt and
blow down system emission factors were taken from the U.S. Inventory (EPA, 2008). Based on
the updated industry profile, U.S. petroleum refineries have onsite GHG emissions of 204.75
million metric tons (mmt) of CO2 equivalents (C02e). When accounting for electricity and
steam purchases, petroleum refineries are also responsible for indirect GHG emissions of 26.0
mmt CC>2e.

Figure 1 presents the breakdown of onsite GHG emissions by source. As seen in Figure 1,
combustion sources, direct process emissions, and flaring account for 99 percent of the onsite
GHG emissions (on a C02e basis).

11


-------
Blowdown
0.18%

Wastewater
T reatment
0.43%

Cooling Towers
0.003%

Asphalt Blowing
0.10%

Delayed Coking
^ 0.058%

^Equipment Leaks
0.014%

\_Storage Tanks
0.007%

Flaring
1.6%

Sulfur Plant
1.9%

H2 Plant
2.7%

Figure 1. Relative importance of refinery GHG emission sources evaluated on COie basis.

3. Emission Sources

This section includes brief descriptions of the sources and process units that generate significant
greenhouse gases at a refinery. More complete descriptions of the entire refining process are
available in other locations (U.S. EPA, 1995; U.S. EPA, 1998; U.S. DOE, 2007).

3.1 Stationary Combustion Sources

As seen in Figure 1, the combustion of fuels in stationary combustion sources is a significant
source of GHG emissions at petroleum refineries. Combustion sources include process heaters,
boilers, combustion turbines, and similar devices. Nearly all refinery process units use process
heaters. In addition to direct process heat, many refinery processes also have steam and
electricity requirements. Some refineries purchase steam to meet their process's steam
requirements; others use dedicated on-site boilers to meet their steam needs. Similarly, some
refineries purchase electricity from the grid to run their pumps and other electrical equipment;
other refineries have co-generation facilities to meet their electricity needs and may produce
excess electricity to sell to the grid. Refineries that produce their own steam or electricity will
have higher on-site fuel usage, all other factors being equal, than refineries that purchase these

12


-------
utilities. Combustion sources primarily emit CO2, but they also emit small amounts of CH4 and
N20.

The predominant fuel used at petroleum refineries is refinery fuel gas (RFG), which is also
known as still gas. RFG is a mixture of light CI to C4 hydrocarbons, hydrogen, hydrogen
sulfide (H2S), and other gases that exit the top (overhead) of the distillation column and remain
uncondensed as they pass through the overhead condenser. RFG produced at different locations
within the refinery is typically compressed, treated to remove H2S (if necessary), and routed to a
common, centralized location (i.e., mix drum) to supply fuel to the various process heaters at the
refinery. This RFG collection and distribution system is referred to as the fuel gas system.

The fuel gas generated at the refinery is typically augmented with natural gas to supply the full
energy needs of the refinery. Depending on the types of crude oil processed and the process
units in operation, the amount of supplemental natural gas needed can change significantly.
Topping and hydroskimming refineries that process heavier crude oils generate limited amounts
of RFG and may use natural gas for 70 percent or more of their energy needs.
Upgrading/conversion refineries, especially those processing lighter crude oils, may need very
little supplemental natural gas or may even produce more fuel gas than needed for basic
operations. Depending on the quantities of propane and butane (C3 and C4 hydrocarbons)
produced and local market conditions, upgrading/conversion refineries will typically have light
gas plant to recover propane and butane for sale as products or for use as petrochemical or gas
blending feedstocks. In these refineries, the hydrocarbon content of the RFG will be dominated
by CI or C2 hydrocarbons. Consequently, there may be significant variability in the fuel gas
composition between different refineries. Within a given refinery, the variability in the refinery
fuel gas composition should be somewhat less because most refinery process units are
continuous. However, delayed coking units, which are significant fuel gas producers, are batch
processes, so refineries with these units may have more variability in RFG composition than
other refineries. Additionally, certain process units may cycle operations (notably hydrotreating
units and catalytic reforming units) and other units are occasionally taken off-line for
maintenance so that occasional variability in fuel gas composition within a refinery is inevitable.

3.2 Flares

Flares are commonly used in refineries as safety devices to receive gases during periods of
process upsets, equipment malfunctions, and unit start-up and shutdowns. "Emergency" flares
receive only low flows of "sweep" gas to prevent air (oxygen) from entering the flare header and
possibly the fuel gas system while maintaining the readiness of the flare in the event of a
significant malfunction or process upset. Some flares may receive excess process gas on a
frequent or routine basis; these flares act as pressure relief systems for the refinery's fuel gas
system and may also be used to combust other low pressure gas streams generated at the
refinery. Some flares may be used solely as control devices for regulatory purposes. As with
stationary combustion sources, the combustion of gas in a flare results predominately in
emissions of CO2 along with small amounts of CH4 and N2O. In the 2002 National Emissions
Inventory (NEI), emissions were reported for 252 flares at 77 refineries (EPA, 2006). Based on
these data, each refinery is expected to have 3 flares on average.

13


-------
3.3	Catalytic Cracking Units

In the catalytic cracking process, heat and pressure are used with a catalyst to break large
hydrocarbons into smaller molecules. The fluid catalytic cracking unit (FCCU) is the most
common type of catalytic cracking unit currently in use. Based on process-specific data
collected in the development of emission standards for petroleum refineries, there are
approximately 125 FCCU in the petroleum refining industry. In this type of reactor, the feed is
pre-heated to between 500 and 800 degrees Fahrenheit (°F) and contacted with fine catalyst
particles from the regenerator section, which are at about 1,300°F in the feed line ("riser"). The
feed vapor, which is heavy distillate oil from the crude or vacuum distillation column, reacts
when contacted with the hot catalyst to break (or crack) the large hydrocarbon compounds into a
variety of lighter hydrocarbons. During this cracking process, coke is deposited on the catalyst
particles, which deactivates the catalyst. The catalyst separates from the reacted ("cracked")
vapors in the reactor; the vapors continue to a fractionation tower and the catalyst is recycled to
the regenerator portion of the FCCU to burn-off the coke deposits and prepare the catalyst for
reuse in the FCCU riser/reactor (U.S. EPA, 1998).

The FCCU catalyst regenerator generates GHG through the combustion of coke, which is
essentially solid carbon with small amounts of hydrogen and various impurities that were
deposited on the catalyst particles during the cracking process. CO2 is the primary GHG emitted;
small quantities of CH4 and N2O are also emitted during "coke burn-off" An FCCU catalyst
regenerator can be designed for complete or partial combustion. A complete-combustion FCCU
operates with sufficient air to convert most of the carbon to CO2 rather than carbon monoxide
(CO). A partial-combustion FCCU generates CO as well as CO2, so most partial-combustion
FCCUs are typically followed by a CO boiler to convert the CO to CO2. Most refineries that
operate an FCCU recover useful heat generated from the combustion of catalyst coke during
catalyst regeneration; the heat recovered from catalyst coke combustion offsets some of the
refinery's ancillary energy needs.

Thermal catalytic cracking units (TCCU) are similar to FCCUs except that the catalyst particles
are much bigger and the system uses a moving bed reactor rather than a fluidized system. The
generation of GHG, however, is the same. Specifically, GHG are generated in the regenerator
section of the TCCU when coke deposited on the catalyst particles is burned-off in order to
restore catalyst activity.

3.4	Catalytic Reforming Units

In the catalytic reforming unit (CRU), low-octane heavy hydrocarbons, generally gasoline and
naptha are reacted with a catalyst to produce aromatic compounds such as benzene. The feed to
the CRU is usually treated with hydrogen to remove sulfur, nitrogen and metallic contaminants.
The CRU usually has a series of three to six fixed-bed or moving bed reactors and may be
operated continuously or as a semi-regenerative unit. As in the FCCU, coke is deposited on
catalyst particles during the processing reaction, and this "catalyst coke" must be burned-off to
reactivate the catalyst, generating CO2, along with small amounts of CH4 and N2O. In a
continuous CRU, the catalyst can be regenerated one reactor at a time, which avoids disrupting

14


-------
the process. In a semi-regenerative CRU, all the reactors are shut down at one time for catalyst
regeneration, usually after no more than 2 years of operation (U.S. EPA, 1998).

3.5	Fluid Coking Units

Coking is another cracking process, usually used at a refinery to generate transportation fuels,
such as gasoline and diesel, from lower-value fuel oils. A desired by-product of the coking
reaction is petroleum coke, which can be used as a fuel for power plants as well as a raw material
for carbon and graphite products. The fluid coking process is continuous and occurs in a reactor
rather than a coke drum like the delayed coking process. Fluid coking units produce a higher
grade of petroleum coke than delayed coking units; however, unlike delayed coking units that
use large process preheaters, fluid coking units burn 15 to 25 percent of the coke produced to
provide the heat needed for the coking reactions (U.S. DOE, 2007). Like the FCCU and CRU,
the combustion of the petroleum coke generates CO2 along with small amounts of CH4 and N2O.

3.6	Sulfur Recovery Vents

Hydrogen sulfide is removed from the refinery fuel gas system through the use of amine
scrubbers. While the selectivity of hydrogen sulfide removal is dependent on the type of amine
solution used, these scrubbers also tend to extract CO2 from the fuel gas. The concentrated sour
gas is then processed in a sulfur recovery plant to convert the hydrogen sulfide into elemental
sulfur or sulfuric acid. CO2 in the sour gas will pass through the sulfur recovery plant and be
released in the final sulfur plant vent. Additionally, small amounts of hydrocarbons may also be
present in the sour gas stream. These hydrocarbons will eventually be converted to CO2 in the
sulfur recovery plant or via tail gas incineration. The most common type of sulfur recovery plant
is the Claus unit, which produces elemental sulfur. The first step in a Claus unit is a burner to
convert one-third of the sour gas into sulfur dioxide prior to the Claus catalytic reactors. GHG
emissions from the fuel fired to the Claus burner are expected to be accounted for as a
combustion source. After that, the sulfur dioxide and unburned hydrogen sulfide are reacted in
the presence of a bauxite catalyst to produce elemental sulfur. Based on process-specific data
collected in the development of emission standards for petroleum refineries, there are 195 sulfur
recovery trains in the petroleum refining industry (U.S. EPA, 1998).

3.7	Hydrogen Plants

The most common method of producing hydrogen at a refinery is the steam methane reforming
(SMR) process. Methane, other light hydrocarbons, and steam are reacted via a nickel catalyst to
produce hydrogen and CO. Excess CH4 is added and combusted to provide the heat needed for
this endothermic reaction. The CO generated by the initial reaction further reacts with the steam
to generate hydrogen and CO2 (U.S. DOE, 2007). According to EIA's Refinery Capacity Report
2006 (EIA, 2006), 54 of the 150 petroleum refineries have hydrogen production capacity.

3.8	Fugitive Emission Sources

Fugitive CH4 emission sources are projected to only contribute 0.8 percent of a typical refinery's
total GHG emissions. The largest four fugitive emission sources are projected to be blow down
systems, delayed coking unit depressurization and coke cutting, asphalt blowing, and wastewater
treatment (including sludge digestion). Methane emissions from process equipment leaks are
expected to be small compared to other GHG emission sources at a typical refinery. Unless the

15


-------
refinery receives unstabilized crude, CH4 emissions from crude oil loading and storage are
expected to be negligible, as are CH4 emissions from cooling towers. These processes are
described in further detail in Appendix A.

4.	Review of Existing Programs and Methodologies

In developing GHG monitoring and reporting options for petroleum refineries, a number
of existing programs and guideline methodologies were reviewed. Specifically, the following
resources were examined:

1.	2006 Intergovernmental Panel on Climate Change (IPCC) Guidelines for National
Greenhouse Gas Inventories. Volume 2, Chapters 2 and 4.

2.	European Union (EU) Emissions Trading Scheme (2007). 2007/589/EC: Commission
Decision of 18 July 2007 Establishing Guidelines for the Monitoring and Reporting of
Greenhouse Gas EmissionsPpursuant to Directive 2003/87/EC of the European
Parliament and of the Council. Available at:

http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CELEX:32007D0589:EN:NQT.

3.	U.S. Department of Energy (DOE). 2007. Technical Guidelines: Voluntary Reporting Of
Greenhouse Gases (1605(B)) Program.

4.	API (American Petroleum Institute). 2004. Compendium of Greenhouse Gas Emissions
Methodologies for the Oil and Gas Industry. February.

5.	CARB (California Air Resource Board). 2008. Regulation For The Mandatory
Reporting of Greenhouse Gas Emissions: Second 15-Day Modified Regulatory
Language For Public Comment. May 15.

6.	Environment Canada (2006). Technical Guidance Manual on Reporting Greenhouse Gas
Emissions. http://www.ghgreporting.gc.ca/GHGInfo/Pages/pagel5.aspx?lang=E.

Each of these sources was reviewed to determine the types of emissions to be reported, the
facility reporting thresholds, and the monitoring methodologies recommended. The reporting
and monitoring options presented in Sections 5, 6, and 7 are commensurate with the
methodologies used in these existing programs and guidelines.

5.	Types of Information to be Reported

5.1 Types of Emissions to be Reported

Based on the existing programs and the emission sources at petroleum refineries, GHG reporting
for refineries are limited to CO2, CH4, and N2O. Table 2 summarizes the refinery emission
sources expected to have appreciable GHG emissions and the GHG expected to be emitted for
each refinery emission source.

16


-------
Figure 2 presents the breakdown of onsite GHG emissions by pollutant. Figure 2 indicates that
CH4 and N2O emissions from petroleum refineries account for less than 1 percent of the total
GHG emissions (evaluated on a C02e basis).

Table 2. Summary of Refinery GHG Emission Sources Considered for Reporting

Emission Source

GHG Emitted

Stationary combustion sources

C02, CH4, and N20

Coke burn-off emissions from catalytic cracking units, fluid
coking units, catalytic reforming units, and coke calcining units

C02, CH4, and N20

Flares

C02, CH4, and N20

Hydrogen plant vent

C02 and CH4

Sulfur recovery plant

C02

On-site wastewater treatment system

C02 and CH4

On-site land disposal unit

ch4

Asphalt blowing

C02 or CH4

Uncontrolled blowdown systems

ch4

Process vents not otherwise specified

C02, CH4, and N20

Delayed coking units

ch4

Process equipment leaks

ch4

Storage tanks

ch4

Loading operations

ch4

17


-------
N20 	 	 CH4

0.09%	0.87%

C02
99.04%

Figure 2. Relative importance of CO2, CH4, and N2O in the total nationwide refinery GHG
emissions inventory (evaluated on a COie basis).

5.2 Other Information to be Reported

In order to check the reported GHG emissions for reasonableness and for other data quality
considerations, additional information about the emission sources is needed. Although the exact
information required is somewhat source dependent, the following is a general list of additional
information that must be reported.

(1)	The unit identification number (if applicable);

(2)	A description of the type of unit (RFG-fired process heater, flare, FCCU, TCCU, sulfur
recovery plant, etc.);

(3)	Maximum rated throughput of the unit (MMBtu/hr; bbl/stream day, tons sulfur
produced/stream day, etc.);

(4)	The calculated CO2, CH4, and N2O emissions for each unit (as applicable), expressed in both
metric tons of pollutant emitted and metric tons of C02e; and

(5)	A description of the method used to measure and/or calculate the GHG emissions for each
unit.

18


-------
6. Options for Reporting Threshold

Several options were evaluated as potential reporting thresholds. These options include:

Option 1. All refineries with facility-wide GHG emissions exceeding 1,000 metric tons of
C02e (mtC02e) must report.

Option 2. All refineries with facility-wide GHG emissions exceeding 10,000 mtC02e must
report.

Option 3. All refineries with facility-wide GHG emissions exceeding 25,000 mtC02e must
report.

Option 4. All refineries with facility-wide GHG emissions exceeding 100,000 mtC02e must
report.

Table 3 presents the number of refineries and the GHG emissions included for each threshold
option. The analysis presented in Table 3 only considers the direct (on-site) GHG emissions at
the refineries. Essentially all refineries are expected to have direct emissions exceeding a
10,000 mtC02e threshold. If indirect GHG emissions are included, all U.S. petroleum refineries
are expected to exceed a 25,000 mtC02e emission threshold.

Table 3. Evaluation of Alternative Threshold Options (considering direct emissions only)

Option/Threshold Level

Emissions Covered

Facilities Covered

mmt C02e/year

Percent

Number

Percent

Option 1: >1,000 mtC02e

204.75

100

150

100

Option 2: >10,000 mtC02e

204.74

99.995

149

99.3

Option 3: >25,000 mtC02e

204.69

99.97

146

97.3

Option 4: >100,000 mtC02e

203.75

99.51

128

85.3

7. Options for Monitoring Methods

7.1 Stationary Combustion Sources

There are four basic monitoring options for combustion units. Option 1 is to use annual fuel
consumption (based on company records), default higher heating value (HHV) for the fuel, and a
fuel-specific emission factor. Because of the variability in refinery fuel gas composition, this
method has high uncertainty when applied to a specific refinery. The carbon content or HHV of
other fuel products used at a refinery, such as diesel fuel or directly purchased and used natural
gas, will have much less variability both within the refinery and across different refineries than
refinery fuel gas. As such, Option 1 can yield reasonably accurate CO2 emission estimates for
these non-RFG fuel types. However, these are expected to be only a small portion of a refinery's
total fuel combustion. Option 2 is similar to Option 1, but the HHV is measured periodically
(daily or weekly measurement frequencies were evaluated) and fuel consumption quantities
(based on company records) are estimated over the same intervals as the HHV measurements.
This option should reduce the uncertainty associated with Option 1 since a higher sampling
frequency yields lower uncertainties. However, since hydrogen combustion does not produce

19


-------
CO2, it is important to note that a measure of HHV does not necessarily improve the estimation
of CO2 emissions since some of the heat content of RFG is from hydrogen and only carbon
content measurements can accurately estimate carbon dioxide emissions for fuel gas streams
with significant hydrogen content. Option 3 uses direct volumetric flow measurements and
routine carbon content measurements (daily or weekly) to estimate CO2 emissions. Like Option

2,	this option will have less uncertainty than Option 1 since a higher sampling frequency yields
lower uncertainties. Option 3 is preferred to Option 2 in that some of the heat content of refinery
fuel gas is from hydrogen (again, only carbon content measurements can accurately estimate CO2
emissions for fuel gas streams with significant hydrogen content). Therefore, routine
measurement of the carbon content provides a better correlation to the resulting CO2 emissions
than HHV. Also, in Option 3 direct fuel flow measurements are made using calibrated flow
meters, whereas Option 2 allows less rigorous estimates of fuel consumption. As a variant to
Option 3, the use of a continuous carbon content monitor was also evaluated. Option 4 uses a
continuous emission monitoring system (CEMS) for CO2 and exhaust flow rate. These options
(or Tiers) are described in more detail in the Stationary Combustion TSD (EPA-HQ-OAR-2008-
0508-004).

7.2	Flares

Three general options were considered for flares, which follow Options 1 through 3 for
stationary combustion sources; as flares do not have enclosed exhaust stacks, an exhaust CEMS
(Option 4) is not technically feasible for flares. Stationary combustion source Options 1 and 2
can provide reasonable estimates of the "sweep" gas or routine flare gas GHG emissions but
cannot be used to provide accurate estimates of the GHG emissions released during periods of
start-up, shutdown, or malfunction (SSM) because the flow rate and composition of the gases
released to the flare during SSM events can vary so widely. As such, Options 1 and 2 for flares
require separate engineering calculation of the GHG emissions from flares that occur during
SSM events. Due to the variability in flow rates and potential gas composition for flares, Option

3,	especially the variant of Option 3 that requires both continuous flow and continuous
composition monitoring, provides significantly more accurate emission estimates than Options 1
or 2; however, this variability places additional requirements on the types of monitors that can be
used for Option 3. Dual range monitors will generally be required to monitor the low flow rates
associated with maintaining the readiness of the flare and the high flow rates that can occur
during periods of process upsets, which adds additional costs to the monitoring system.

7.3	Process Emissions

Four processes are considered in this section; a fifth process, hydrogen production by steam
reforming, is considered in a separate Hydrogen Production TSD (EPA-HQ-OAR-2008-0508-
016). Three of the processes considered here, the FCCU, the catalytic reforming unit (CRU),
and the fluid coking unit vents, are all associated with petroleum coke combustion. As described
previously, the emissions generated from these processes are the result of coke combustion, so
the monitoring methods are similar to those in the combustion section, but they are tailored by
process considering the relative quantity of coke burned by each process and the typical
measurement locations and controls used for each type of process.

20


-------
Three monitoring options were considered for these process emissions sources. Option 1 is to
use process feed rates and default coke burn-off factors and coke carbon content to estimate CO2
emissions. This option does not account for differences in coke deposition rates by type of crude
or severity of the cracking conditions used by a given unit, and therefore is expected to have high
uncertainties. Option 2 is to monitor coke burn-off rates using gas composition analyzers and to
estimate flow rate by air blast rates and flow correlations. This option is required for many
FCCU to comply with Part 60 and Part 63 standards and provides accurate estimates of the CO2
emissions rate from the FCCU regenerator. Due to these regulatory requirements, it is estimated
that approximately 90 percent of FCCUs already have gas compositional monitors in-place.
Option 3 is the use of a CEMS for CO2 and exhaust gas flow rate. For partial combustion units,
this may lead to additional monitoring needs to prevent double counting the GHG process
emissions from the FCCU and GHG combustion emissions from RFG fired in the CO boiler.

A fourth process vent source is the sulfur recovery vent. As described previously, some CO2 and
trace amount of hydrocarbons will enter the sulfur recovery plant in the sour gas. It is expected
that the total sour gas flow rate is metered at the plant as it is necessary for proper operation of
the sulfur recovery plant. Option 1 entails periodic sampling of the sour gas for carbon content
at the inlet to the sulfur recovery plant. Option 2 is the use of a CEMS capable of measuring the
carbon content and flow rate in the sour gas feed stream at the inlet to the sulfur recovery plant.
Option 2 would have less uncertainty than Option 1 due to the higher measurement frequency.
Option 3 is the use of a CEMS for measuring the CO2 and flow rate of the final sulfur recovery
plant tail gas stack. Option 3 has the advantage of including the CO2 emissions from the first-
stage Claus burner as well as the tail gas incinerator, if one is used. However, this option could
result in double counting because fuel used in the Claus burner would also contribute to the CO2
emissions in the final tail gas stack. If a refinery is using "common-pipe" fuel gas consumption
measurements, it will have no way of knowing how much of the exhaust CO2 is from the
combustion of fuel gas in the Claus burner. In this case, the refinery would have to measure fuel
flow and composition directly at the Claus burner in addition to the stack CEMS measurements.
Given the issues related in properly accounting for the sulfur plant process emissions and
combustion emissions, Option 3 may not provide any improvement in the overall accuracy of the
process-related GHG emissions estimate for sulfur recovery plants.

7.4 Fugitive Emission Sources

There are a variety of GHG emission sources at the refinery, which include: asphalt blowing,
delayed coking unit depressurization and coke cutting, coke calcining, blowdown systems,
process vents, process equipment leaks, storage tanks, loading operations, wastewater treatment,
and waste disposal. To fully account for the refinery's GHG emissions, the emissions from these
sources should be reported; however, the emissions from these sources are expected to be only
about 1 percent of the refinery's total GHG emissions. Therefore, Option 1 - use of default
emission factors - is appropriate for these sources. Where applicable, more stringent options
(such as using Method 21 monitoring for process equipment leaks, using the TANKS model and
compositional data for storage tanks, and using the number of delayed coking unit vessel size
and cycle activity data) were also considered. If a refinery is already monitoring high CH4-
containing process lines, or is already using the TANKS model to estimate methane emissions
from storage tanks, etc., then these options can be used. For purposes of the current regulatory

21


-------
activity, additional monitoring of these fugitive CH4 emission sources is not specified given the
high cost and high uncertainty associated with the fugitive monitoring or estimation methods and
the small contribution these emissions make to the total GHG emissions from the refinery.

8.	Options for Estimating Missing Data

A complete record of all measured parameters used in the GHG emissions calculations is
required (e.g., concentrations, flow rates, fuel heating values, carbon content values, etc.).
Therefore, whenever a quality-assured value of a required parameter is unavailable (e.g., if a
CEMS malfunctions during unit operation or if a required fuel sample is not taken), a substitute
data value for the missing parameter shall be used in the calculations.

In general, it is recommended that the average of the data measurements before and after the
missing data period be used to calculate the emissions during the missing data period. If, for a
particular parameter, no quality-assured data are available prior to the missing data incident, the
substitute data value should be the first quality-assured value obtained after the missing data
period. Missing data procedures are applicable for heat content, carbon content, gas and liquid
fuel flow rates, stack gas or air blast flow rates, and compositional analysis data (CO2, CO, O2,
CH4, N2O, and H20 content, as applicable).

9.	QA/QC Requirements

To ensure the quality of the reported GHG emissions, the following quality assurance/quality
control (QA/QC) activities are considered important:

(1)	Developing and maintaining a Quality Assurance Project Plan (QAPP) that documents the
measurements made, their accuracy, and explains the quality assurance procedures applied
for each measurement used to quantify GHG emissions. The QAPP and the appropriate
records of quality assurance checks should be retained on-site for a minimum of 5 years.

(2)	All fuel flow meters, gas composition monitors, and/or heating value monitors that are used
to provide data for the GHG emissions calculations should be calibrated prior to the first
reporting year, using a suitable method published by a consensus standards organization
(e.g., ASTM, ASME, API, AGA, etc.). Alternatively, calibration procedures specified by
the flow meter manufacturer may be used. Fuel flow meters, gas composition monitors,
and/or heating value monitors shall be recalibrated either annually or at the minimum
frequency specified by the manufacturer.

(3)	Documentation of the procedures used to ensure the accuracy of the estimates of fuel usage,
gas composition, and/or heating value including, but not limited to, calibration of weighing
equipment, fuel flow meters, and other measurement devices should maintained. The
estimated accuracy of measurements made with these devices should also be recorded, and
the technical basis for the estimates should be provided.

(4)	All CO2 CEMS and flow rate monitors used for direct measurement of GHG emissions
should comply with QA procedures for daily calibration drift checks and quarterly or annual

22


-------
accuracy assessments, such as those provided in Appendix F to Part 60 or similar QA
procedures.

10. References

API (American Petroleum Institute). 2004. Compendium of Greenhouse Gas Emissions
Methodologies for the Oil and Gas Industry. February.

CARB (California Air Resource Board). 2008. Regulation For The Mandatory Reporting of
Greenhouse Gas Emissions: Second 15-Day Modified Regulatory Language For Public
Comment. Available at: http://www.arb.ca.gov/regact/2007/ghg2007/ghgattachment 1.pdf.
May 15.

Coburn, J., 2007. Greenhouse Gas Industry Profile for the Petroleum Refining Industry.

Prepared for Lisa Hanle, U.S. Environmental Protection Agency, Washington. June 11.

EI A (Energy Information Administration). 2006a. Petroleum Supply Annual 2005. Prepared by
the Energy Information Administration, Washington, DC. October 23.

EIA (Energy Information Administration). 2006b. Refinery Capacity Report 2006. Prepared by
the Energy Information Administration, Washington, DC. June 15.

Environment Canada (2006). Greenhouse Gas Emissions Reporting: Technical Guidance on
Reporting Greenhouse Gas Emissions. Available at:
http://www.ghgreporting.gc.ca/GHGInfo/Pages/pagel 5.aspx?lang=E.

European Union (EU) Emissions Trading Scheme. 2007. 2007/589/EC: Commission Decision
of 18 July 2007 Establishing Guidelines for the Monitoring and Reporting of Greenhouse
Gas Emissions Pursuant to Directive 2003/87/EC of the European Parliament and of the
Council. Available at: http://eur-

lex.europa.eu/LexUriServ/LexUriServ.do?uri=CELEX:32007D0589:EN:NQT. July.

IPCC. 2006. 2006 Intergovernmental Panel on Climate Change (IPCC) Guidelines for National
Greenhouse Gas Inventories. Volume 2, Energy; Chapters 2 and 4. Available at:
http://www.ipcc-nggip.iges.or.ip/public/2006gl/vol2.html.

U.S. Department of Energy (DOE). Technical Guidelines: Voluntary Reporting Of Greenhouse
Gases (1605(B)) Program. January 2007.

U.S. Department of Energy. 2007. Energy and Environmental Profile of the U.S. Petroleum
Refining Industry. Prepared by Energetics, Inc., Columbia, MD. November.

U.S. Environmental Protection Agency. 1995. EPA Office of Compliance Sector Notebook

Project: Profile of the Petroleum Refining Industry. EPA/310-R-95-013. Washington,
DC, September.

23


-------
U.S. Environmental Protection Agency. 1998. Petroleum Refineries—Background Information
for Proposed Standards, Catalytic Cracking (Fluid and Other) Units, Catalytic Reforming
Units, and Sulfur Recovery Units. EPA-453/R-98-003. Washington, DC: Government
Printing Office.

U.S. Environmental Protection Agency. 2006. Documentation for the Final 2002 Point Source
National Emissions Inventory. Research Triangle Park, NC. February 10. Report and
data available at: http://www.epa.gov/ttn/chief/net/2002inventorv.html.

U.S. Environmental Protection Agency. 2008. Inventory of Greenhouse Gas Emissions and

Sinks: 1990-2006. EPA-430-R-08-005. Office of Atmospheric Programs, Washington,
DC. April 15.

24


-------
Appendix A. Derivation of Calculation Methods

Most of the calculation methods being considered for the reporting of GHG emissions from
refineries are fairly straight-forward and are documented elsewhere.1'2 This section provides
additional information for a few of the methods being evaluated and were specifically developed
to reduce the burden for estimating fugitive emission from sources that are not expected to
contribute significantly to the overall GHG inventory for U.S. petroleum refineries.

A. 1 Equipment Leaks

The Equipment Leaks Protocol Document contains a variety of methods that may be used to
estimate fugitive emissions from leaking process equipment.3 To use these methods for
estimating methane emissions, the average methane concentration of the various refinery process
streams would need to be measured or estimated, and the number of equipment components
would need to be determined. Many refineries already have these data, and can use the methods
in the Protocol Document directly. For refineries that do not have these data readily available, a
simple method of estimating these emissions was developed. First, methane emissions from
fugitive equipment leaks were estimated using:

¦	A set leak fraction of 2 percent (1.4 percent with leaks greater than 10,000 ppmv but less
than 100,000 ppmv, and 0.6 percent with leaks greater than 10,000 ppmv).

¦	The zero and pegged value leak equations from the Protocol Document as provided in
Table A-l of this report.

¦	Model refinery equipment component counts from EPA's Locating and Estimating
Emission of Benzene document as well as estimated component counts for fuel gas
systems as provided in Tables A-2 and A-3 of this report.4

¦	An estimated methane composition for each model process unit as provided in Table A-4.
Table A-4 also summarizes the methane emissions per model process unit.

1	EPA, 2008. Inventory of Greenhouse Gas Emissions and Sinks: 1990-2006. EPA-430-R-08-005. Office of

Atmospheric Programs, Washington, DC. April 15.

2	API, 2004. Compendium of Greenhouse Gas Emissions Methodologies for the Oil and Gas Industry. February.

3	U.S. EPA (Environmental Protection Agency). 1995. Protocol for Equipment Leak Emission Estimates. EPA-

453/R-95-017. Office of Air Quality Planning and Standards, Research Triangle Park, NC.

4	U.S. EPA (Environmental Protection Agency). 1998a. Locating and Estimating Air Emissions from Sources of

Benzene. EPA-454/R-98-011. Office of Air Quality Planning and Standards, Research Triangle Park, NC.

25


-------
Table A-l. Fugitive Equipment Leak Rate for Refinery Equipment Components."

Equipment Type (All

Default Zero Emission

Pegged Emission Rates (kg/hr/source)

Services)

Rate (kg/hr/source)

10,000 ppmv

100,000 ppmv

Valve

7.8E-06

0.064

0.140

Pump

2.4E-05

0.074

0.160

Otherb

4.0E-06

0.073

0.110

Connector

7.5E-06

0.028

0.030

Flange

3.1E-07

0.085

0.084

Open-Ended Line

2.0E-06

0.030

0.079

a As reported in U.S. EPA (1995, see footnote 3)

b The "other" equipment type was developed from instruments, loading arms, pressure relief devices, stuffing boxes,
vents, compressors, dump lever arms, diaphragms, drains, hatches, meters, and polished rods. This "other"
equipment type should be applied to any equipment other than connectors, flanges, open-ended lines, pumps, or
valves.

As seen by the estimated emissions for the model process units in Table A-4, the fuel gas system
and the hydrogen plant are expected to produce the majority of the refinery's methane emissions.
From the results presented in Table A-4, the average methane emissions between the small and
large model plants for fuel gas systems and hydrogen plants is 6.0 and 4.3 mt/yr, respectively.
For crude oil distillation columns, the average methane emissions rate is approximately 0.4
mt/yr. Several process units in Table A-4 were projected to have average emission rates of
approximately 0.2 mt/yr, and several other process units were projected to have average emission
rates of approximately 0.1 mt/yr.

Based on this analysis, Equation 1 was developed to provide a very quick and simple method for
estimating methane emissions from fugitive equipment leaks at a refinery.

CH4 = (0.4 * Ncd + 0.2 * Npuj +0.1* NPU2 + 4.3 * NH2 + 6 * NPGS )	(Equation 1)

Where:

CH4 = annual methane emissions from fugitive equipment leaks (mtCH4/yr)

Ncd = number of atmospheric crude oil distillation columns at the facility
Npui = cumulative number of catalytic cracking units, coking units (delayed or fluid),
hydrocracking, and full-range distillation columns (including depropanizer and
debutanizer distillation columns) at the facility
Npu2 = cumulative number of hydrotreating/hydrorefining units, catalytic reforming units, and

visbreaking units at the facility
Nh2 = total number of hydrogen plants at the facility
Nfgs = total number of fuel gas systems at the facility

26


-------
Table A-2. Median Equipment Leak Component Counts for Small Model Processes."

Process Unit

Valves

Pumps

Com-
pres-
sors

Pressure Relief
Valves

Flanges

Open- Sampling

Gas

Light
Liquid

Heavy
Liquid

Light
Liquid

Heavy
Liquid

Gas

Light
Liquid

Heavy
Liquid

Gas

Light
Liquid

Heavy
Liquid

Ended
Lines

Connec-
tions

Crude Distillation

75

251

216

8

8

2

6

6

5

164

555

454

39

10

Alkylation (sulfuric acid)

278

582

34

18

10

1

12

15

4

705

1296

785

20

16

Alkylation (HF)

102

402

62

13

3

2

12

13

0

300

1200

468

26

8

Catalytic Reforming

138

234

293

8

5

3

5

3

3

345

566

732

27

6

Hydrocracking

300

375

306

12

9

2

9

4

4

1038

892

623

25

10

Hydrotreating/Hydrorefining

100

208

218

5

5

2

5

3

5

290

456

538

20

6

Catalytic Cracking

186

375

450

13

14

2

8

8

7

490

943

938

8

8

Thermal Cracking
(visbreaking)

206

197

0

7

0

0

4

0

0

515

405

0

0

4

Thermal Cracking (coking)

148

174

277

9

8

2

7

16

13

260

322

459

13

8

Hydrogen Plant

168

41

0

3

0

2

4

2

0

304

78

0

8

4

Asphalt Plant

120

334

250

5

8

2

5

10

9

187

476

900

16

6

Product Blending

67

205

202

6

11

1

10

6

22

230

398

341

33

14

Sulfur Plant

58

96

127

6

6

3

3

88

15

165

240

345

50

3

Vacuum Distillation

54

26

84

6

6

2

2

5

2

105

121

230

16

4

Full-Range Distillation

157

313

118

7

4

2

5

4

6

171

481

210

20

6

Isomerization

270

352

64

9

2

2

7

10

1

432

971

243

7

8

Polymerization

224

563

15

12

0

1

10

5

3

150

450

27

5

7

MEK Dewaxing

145

1208

200

35

39

3

10

14

4

452

1486

2645

19

17

Other Lube Oil Processes

153

242

201

7

5

2

5

5

5

167

307

249

60

6

Fuel Gas Systemb

120

0

0

0

0

2

5

0

0

300

0

0

20

5

a Process component counts (except for fuel gas system) as presented in the Benzene L&E document (U.S. EPA, 1998; see footnote 4) for refineries with crude
capacities less than 50,000 bbl/cd

b Fuel gas system component counts estimated using engineering judgment.

27


-------
Table A-3. Median Equipment Leak Component Counts for Large Model Processes."

Process Unit

Valves

Pumps

Com-
pres-
sors

Pressure Relief
Valves

Flanges

Open- Sampling

Gas

Light
Liquid

Heavy
Liquid

Light
Liquid

Heavy
Liquid

Gas

Light
Liquid

Heavy
Liquid

Gas

Light
Liquid

Heavy
Liquid

Ended
Lines

Connec-
tions

Crude Distillation

204

440

498

15

14

2

7

5

12

549

982

1046

75

9

Alkylation (sulfuric acid)

192

597

0

21

0

2

13

4

0

491

1328

600

35

6

Alkylation (HF)

104

624

128

13

8

1

9

11

1

330

1300

180

40

14

Catalytic Reforming

310

383

84

12

2

3

8

11

0

653

842

132

48

9

Hydrocracking

290

651

308

22

12

2

10

12

0

418

1361

507

329

28

Hydrotreating/Hydrorefining

224

253

200

7

6

2

9

4

8

439

581

481

49

8

Catalytic Cracking

277

282

445

12

12

2

11

9

13

593

747

890

59

15

Thermal Cracking
(visbreaking)

110

246

130

7

6

1

6

3

15

277

563

468

30

7

Thermal Cracking (coking)

190

309

250

12

11

1

8

5

10

627

748

791

100

10

Hydrogen Plant

301

58

0

7

360

3

4

139

0

162

148

0

59

21

Asphalt Plant

76

43

0

4

0

0

3

7

0

90

90

0

24

24

Product Blending

75

419

186

10

10

2

9

16

6

227

664

473

24

8

Sulfur Plant

100

125

110

8

3

1

4

4

4

280

460

179

22

7

Vacuum Distillation

229

108

447

2

12

1

5

1

4

473

136

1072

0

7

Full-Range Distillation

160

561

73

14

2

2

7

8

2

562

1386

288

54

6

Isomerization

164

300

78

9

5

2

15

5

2

300

540

265

36

7

Polymerization

129

351

82

6

2

0

7

12

28

404

575

170

17

9

MEK Dewaxing

419

1075

130

29

10

4

33

6

18

1676

3870

468

0

7

Other Lube Oil Processes

109

188

375

5

16

3

8

6

20

180

187

1260

18

9

Fuel Gas Systemb

120

0

0

0

0

2

5

0

0

300

0

0

20

5

a Process component counts (except for fuel gas system) as presented in the Benzene L&E document (U..S. EPA, 1998; see footnote 4) for refineries with crude
capacities of 50,000 bbl/cd of greater.

b Fuel gas system component counts estimated using engineering judgment.

28


-------
Table A-4. Estimated Methane Concentration and Fugitive Emission Rates for Model

Refinery Process Units.

Process Unit

Concentration of Methane
(weight fraction)3

Methane Emission Rate
(tonnes CH4/yr)

Gas Stream

Light Liquid
Stream

Small Model
Plant

Large Model
Plant

Crude Distillation

0.05

0.000005

0.185

0.570

Alkylation (sulfuric acid)

0

0

0

0

Alkylation (HF)

0

0

0

0

Catalytic Reforming

0.01

0

0.074

0.146

Hydrocracking

0.02

0

0.403

0.216

Hydrotreating/hydrorefining

0.01

0

0.059

0.101

Catalytic Cracking

0.02

0

0.205

0.265

Thermal Cracking (visbreaking)

0.01

0

0.109

0.059

Thermal Cracking (coking)

0.02

0

0.125

0.247

Hydrogen Plant

0.6

0.0001

4.300

4.264

Asphalt Plant

0

0

0

0

Product Blending

0

0

0

0

Sulfur Plant

0.001

0

0.003

0.006

Vacuum Distillation

0.001

0

0.002

0.011

Full-Range Distillation

0.02

0

0.101

0.218

Isomerizaiton

0

0

0

0

Polymerization

0

0

0

0

MEK Dewaxing

0

0

0

0

Other Lube Oil Processing

0

0

0

0

Aromatics

0

0

0

0

Fuel Gas System

0.7

0

4.472

7.468

a Methane concentrations for heavy liquids were assumed to be negligible.

A. 2 Storage Tanks

Methane (CH4) emissions can occur from petroleum refinery storage tanks. Crude oil storage
tanks are expected to be the primary contributor to the GHG emissions from storage tanks. Most
other intermediate and final product storage tanks are expected to have negligible CH4 emissions
as the stored liquids typically do not contain any CH4. When crude oil is initially pumped from
the well, the crude oil can contain significant amounts of CH4 because the crude is stored under
pressure within the oil reservoir. When the oil is first stored at atmospheric conditions at the
well site, significant amounts of light organics, including CH4, are released from the crude oil,
commonly referred to as flashing losses. After the light volatiles have flashed from the crude oil,
the "stabilized" crude oil is then transported to the refineries for further processing. As such,
most of the CH4 emissions from crude oil storage typically occur upstream of the petroleum

29


-------
refinery. If the crude oil is transported under pressure (e.g., via pipeline) before the crude oil is
stabilized, significant CH4 emissions can occur at the refinery from flashing losses in the crude
oil storage tanks.

The TANKS model is a tool that implements the AP-42 emission estimation methods (developed
by API) for organic liquid storage.5 The TANKS model is not applicable for estimating the
emissions from unstabilized crude oil. The API Compendium of Greenhouse Gas Emissions
Methodologies for the Oil and Gas Industry contains a variety of methods that may be used to
estimate flashing losses from storage tanks.6 One of the simpler correlation equation methods
presented in the API Compendium and one of the few that is applicable to refinery storage tanks
is referred to as the EUB (Energy and Utilities Board) Rule-of-Thumb approach. The EUB
Rule-of-Thumb correlation equation estimates the volume of gas released. Using the molar
volume of gas at standard conditions and the CH4 content of the gas, the EUB Rule-of-Thumb
approach can be used to calculate flashing losses from storage tanks as follows:

CH4 = (0.995 *Qun *AP)*MFCH4 *-^^-*0.001	(Equation 2)

Where:

CH4 = emission rate of methane from storage tanks (mtCH4/yr)

Qun = quantity of unstabilized crude oil received at the facility (bbl/yr)

AP = pressure differential from the previous storage pressure to atmospheric pressure
(pounds per square inch)

MFch4 = mole fraction of CH4 in vent gas from the unstabilized crude oil storage tank from
facility measurements; use 0.27 as a default if measurement data are not available.

0.995 = correlation equation factor (scf gas per bbl per psi)

16 = molecular weight of CH4 (kg/kg-mole).

MVC = molar volume conversion factor (849.5 scf/kg-mole).

0.001= conversion factor (mt/kg).

For stabilized crude oil, the TANKS model (current version is 4.09D) can be used to estimate
CH4 emissions from crude oil storage tanks. The TANKS model generally outputs total
hydrocarbon losses, so the CH4 content of the released vapor must be determined. It is important
not to use the liquid phase composition for this estimate as the vapors will have much higher
relative CH4 concentrations than in the liquid phase. The TANKS model was used to estimate
the hydrocarbon losses for crude oil using the default crude oil properties pre-loaded in the
TANKS model (Crude oil with a Reid vapor pressure of 5 psi and average molecular weight of
vapor of 50 g/mol). Concentrations of C5 through C8 hydrocarbons in the crude oil were
estimated based on average crude oil compositions reported in Potter and Simmons.7

5	U.S. EPA (Environmental Protection Agency). 1995a. Compilation of Air Pollutant Emission Factors.

Sections 7. AP-42. Office of Air Quality Planning and Standards, Research Triangle Park, NC.

6	API, 2004. Compendium of Greenhouse Gas Emissions Methodologies for the Oil and Gas Industry. February.

7	Potter, T.L., and K.E. Simmons. 1998. Composition of Petroleum Mixtures. Total Petroleum Hydrocarbon Criteria

Working Group Series, Volume 2. May.

30


-------
Concentrations of CI through C4 hydrocarbons in the crude oil were estimated based on best
engineering judgment and adjusted so that the average molecular weight of the equilibrium gas
mixture was 50 g/mol (to match the TANKS model assumption). Although the estimated CH4
concentration in the crude oil was only 5 parts per million by weight (ppmw), the vapor phase
was projected to contain just under 8 percent CH4. Therefore the total hydrocarbon emissions
rates estimated using the TANKS model were multiplied by 8 percent to estimate a CH4
emissions factor for crude oil storage tanks. Several crude oil storage tanks with different
floating roof seal designs and component fittings were evaluated. All of the tanks were 200 ft
diameter, 7.5-million gallon capacity tanks, with an estimated throughput of 20,000 to
25,000 bbls per day (40 to 50 turnovers per year). The tanks were evaluated at a number of
meteorological locations. Based on this evaluation, an emission factor of 0.1 Mg CH4/million
bbls of crude throughput was developed. The uncertainty of this default emission factor is large,
roughly factor of 2 or 3, based on the results for different tank runs. Nonetheless, it provides a
very simple means of estimating CH4 emissions from crude oil storage tanks, and presumably all
storage tanks at the refinery, and is appropriate given the small contribution of storage tank
emissions in the overall GHG emissions inventory for a petroleum refinery.

A. 3 Delayed Coking Unit Coke Cutting

Methane emissions have been found in recent delayed coking vessel depressurization vent tests,
and the tests noted that significant emissions appeared to occur when the coke vessel was
opened.8'9'10'11 Additionally, other test data indicate that significant emissions continue to occur
during the coke cutting operations.12 From the delayed coking vessel depressurization vent tests,
methane is approximately 30 percent of the gas volume on a dry basis. However, water is added
to the coking drum to cool the coke, and the water vaporizes. Consequently, the coker
depressurization vent stream is 90 percent or more water vapor. It is unknown how much of the
void space within the coke drum is well purged with steam and what fraction is trapped and only
released to the atmosphere during the coke cutting operations. Any trapped gas is expected to
have high concentrations of methane (roughly 30 percent by volume).

8	South Coast Air Quality Management District. 2004a. Source Test Report 03-194 Conducted at Chevron / Texaco

Refinery, El Segundo, California—Volatile Organic Compound (VOC), Carbon Monoxide (CO), and Particulate
Matter (PM) Emissions From a Coke Drum Steam Vent. May 14.

9	South Coast Air Quality Management District. 2004b. Source Test Report 03-197 Conducted at Conoco-Phillips

Refinery, Carson, California—Volatile Organic Compound (VOC), Carbon Monoxide (CO), and
Particulate Matter (PM) Emissions From a Coke Drum Steam Vent. July 23.

10	South Coast Air Quality Management District. 2004c. Source Test Report 03-198 Conducted at Exxon Mobil
Refinery, Torrance, California—Volatile Organic Compound (VOC), Carbon Monoxide (CO), and Particulate
Matter (PM) Emissions From a Coke Drum Steam Vent. March 4.

11	South Coast Air Quality Management District. 2004d. Source Test Report 03-200 Conducted at Shell Oil

Refinery, Wilmington, California—Volatile Organic Compound (VOC), Carbon Monoxide (CO), and
Particulate Matter (PM) Emissions From a Coke Drum Steam Vent. July 1.

12	Chambers, A., and M. Strosher. 2006. Refinery Demonstration of Optical Technologies for Measurement of

Fugitive Emissions and for Leak Detection. Prepared for Environment Canada, Ontario Ministry of the
Environment, and Alberta Environment. Project No. CEM 9643-2006. March 31.

31


-------
Although some of the coke drum is filled with coke, the coke is quite porous, so a rough estimate
of the gas within the coke drum can be estimated by the dimensions of the coke drum vessel
itself. A default methane concentration of 3 percent by volume is recommended based on the
typical methane dry composition of 30 percent by volume and the lower range of water content
in the depressurization purge vent. If a facility has data for the coker drum purge vent (but not
during periods of active steaming), then that concentration can be used to estimate the vapor-
phase concentration within the coker drum. While using the direct volume of the coke drum
vessel will over-estimate the amount of gas released, the actual concentration of the gas within
the vessel is expected to be somewhat higher than the purge vent concentrations because of the
likelihood of trapped gas pockets. Thus, the combination of the purge vent concentration
combined with the entire vessel volume assumption is expected to provide a reasonable estimate
of the methane emissions when the coke drum vessel is opened and subsequently de-coked.
Consequently, Equation 3 was developed to estimate the methane emissions from opening the
coke drum vessel and subsequent coke cutting operations.

CH4 =

C	7r * ]~)2 16	^

N*H*	*	*MFch* 0.001

v	4 MVC CH4 j

(Equation 3)

Where:

CH4 = annual CH4 emissions from the delayed coking unit vessel opening (mtCH4/yr)
N = total number of vessel openings for all delayed coking unit vessels of the same

dimensions during the year
H = height of coking unit vessel (ft)

D = diameter of coking unit vessel (ft)

16 = molecular weight of CH4 (kg/kg-mole)

MVC = molar volume conversion (849.5 scf/ kg-mole)

MFch4 = mole fraction of methane in coking vessel gas; default value is 0.03
0.001 = conversion factor - kg to metric tonnes.

32


-------